Normal Mode Analysis As a Routine Part of a Structural Investigation
Total Page：16
File Type：pdf, Size：1020Kb
molecules Review Normal Mode Analysis as a Routine Part of a Structural Investigation Jacob A. Bauer * , Jelena Pavlovi´cand Vladena Bauerová-Hlinková Institute of Molecular Biology, Slovak Academy of Sciences, Dúbravská cesta 21, 845 51 Bratislava, Slovakia * Correspondence: [email protected]; Tel.: +421-2-5930-7439 Received: 29 July 2019; Accepted: 30 August 2019; Published: 10 September 2019 Abstract: Normal mode analysis (NMA) is a technique that can be used to describe the flexible states accessible to a protein about an equilibrium position. These states have been shown repeatedly to have functional significance. NMA is probably the least computationally expensive method for studying the dynamics of macromolecules, and advances in computer technology and algorithms for calculating normal modes over the last 20 years have made it nearly trivial for all but the largest systems. Despite this, it is still uncommon for NMA to be used as a component of the analysis of a structural study. In this review, we will describe NMA, outline its advantages and limitations, explain what can and cannot be learned from it, and address some criticisms and concerns that have been voiced about it. We will then review the most commonly used techniques for reducing the computational cost of this method and identify the web services making use of these methods. We will illustrate several of their possible uses with recent examples from the literature. We conclude by recommending that NMA become one of the standard tools employed in any structural study. Keywords: normal mode analysis; X-ray crystallography; crystal structure; protein dynamics; elastic network model 1. Introduction This special issue contains many articles describing several of the exciting new developments that have occurred in X-ray crystallography in recent years. In this review, we would like, however, to call attention to an old, well-established, but underused technique—Normal Mode Analysis (NMA). NMA is a technique that can be used to describe the flexible states accessible to a protein about an equilibrium position. These states have been shown repeatedly to have functional significance, and there have been many reviews dedicated to the applications of NMA to the study of biological macromolecules [1–9]. At least one book has also been written on this topic [10]. NMA is probably the least computationally expensive method for studying the dynamics of macromolecules, and advances in computer technology and algorithms for calculating normal modes over the last 20 years have made it nearly trivial for all but the largest systems. Despite this, it is still uncommon for NMA to be used as a component of the analysis of a structural study. For example, the number of Web of Science citations (excluding reviews) for the most commonly used NMA software over the last 20 years is only 1757, while the PDB grew by more than 141,000 structures during the same time period. Even allowing that a single study typically produces multiple structures, this is still an enormous discrepancy. What makes this situation even more surprising is that NMA in some form has either been used as or suggested to be used as a tool for aiding molecular replacement [11,12] or for model refinement [12–18] since at least the early 1990s. It has also been used to help fit models into low-resolution cryo-EM maps [19–21] and interpret small-angle X-ray scattering (SAXS) results [22,23]. While it is true that, in earlier years, the equipment and expertise needed for X-ray crystallography and NMA were rarely Molecules 2019, 24, 3293; doi:10.3390/molecules24183293 www.mdpi.com/journal/molecules Molecules 2019, 24, 3293 2 of 20 found together, thereby necessitating separate studies, this no longer needs to be the case. There is no reason why NMA should not become a standard part of the analysis of any newly determined structure. In this review, we will describe NMA, outline its advantages and limitations, explain what can and cannot be learned from it, and address some criticisms and concerns that have been voiced about it. We will then review the most commonly used techniques for reducing the computational cost of this method and identify the web services making use of these methods. Finally, we will review a few select examples from the literature where NMA was used as part of a structural study to answer questions of dynamics by way of illustration, and we conclude with our recommendations. It should be noted at the beginning that our goal is to show how NMA can be useful for a typical structural study rather than to review all the applications of NMA to X-ray crystallography or to highlight the most impressive applications of NMA. 2. Normal Mode Analysis Normal mode analysis is a technique that can be used to describe the flexible states accessible to a protein or other molecule about an equilibrium position. It is based on the physics used to describe small oscillations, which can be found in any book on classical mechanics [24]. The idea is that, when an oscillating system at equilibrium, for example, a protein in an energy minimum conformation, is slightly perturbed, a restoring force acts to bring the perturbed system back to its equilibrium configuration (the minimum energy conformation in the case of protein structures). A system is defined to be in equilibrium or at the bottom of a potential minimum when the generalized forces acting on the system are equal to zero. At the minimum energy conformation, represented by the generalized coordinates q0, the potential energy equation can be represented as a power series in q: 0 0 2 ! 0 ¶V 1 ¶ V V(q) = V(q ) + hi + hihj + . , (1) ¶qi 2 ¶qi¶qj where qi and qj represent the instantaneous configuration of components i and j and the deviation 0 of component i from its equilibrium configuration is given by hi = qi − qi . Superscripts of 0 indicate the equilibrium conformation. V(q) is the potential energy equation of the system and, for proteins, usually takes the form of one of the commonly used molecular dynamics force fields (an overview of these may be found in [25,26]). The first term in the series represents the minimum value of the potential and may be set to zero. The second term will be zero at any local minimum, so the potential can therefore be written !0 1 ¶2V 1 V(q) = hihj = hiVijhj, (2) 2 ¶qi¶qj 2 where Vij is the Hessian matrix that contains the second derivatives of the potential with respect to the components of the system. The Hessian matrix contains information about how changes in the position of one component of the system are tied to changes in the others. It is also necessary to consider the kinetic energy (T) of the system since we are interested in dynamics. For component i, this can be given by 1 d2h T(q) = M i , (3) 2 dt2 where M is a diagonal matrix containing the mass of each particle. The entire equation of motion can be written as 1 d2h 1 M i + h V h = 0. (4) 2 dt2 2 i ij j Molecules 2019, 24, 3293 3 of 20 One solution of this equation is the oscillatory equation hi = aik cos(wkt + dk), (5) where aik is the amplitude of oscillation, wk is the frequency, and dk is a phase factor. By substituting this into Equation (4), the equation of motion can be rewritten into a standard eigenvalue equation: VA = lA, (6) where the matrix A contains the Ak eigenvectors of the Hessian matrix V and l is a diagonal matrix containing the lk eigenvalues. The Ak eigenvectors are the normal mode vectors and describe in which direction and how far each particle in the system moves with respect to the others; the lk eigenvalues equal the squares of the frequencies with which all particles involved in each mode vibrate. Each eigenvector describes in which direction each particle moves and how far it moves with respect to all the other particles in the system, but it does not provide absolute displacements, meaning that simple NMA cannot generally be used to provide the displacement amplitudes of a given normal mode [7]. Generally, the vibrational energy of the system is equally divided across all modes such that the average oscillation amplitude of mode k scales at 1/wk. Modes with higher frequencies, however, have energetically more expensive displacements than those with lower ones. Together, this means that the system experiences the greatest displacements along the lowest frequency or slowest modes: higher frequency modes typically describe rapid local motions, while low frequency modes describe slower collective ones or large-scale conformational changes. Assuming that the system is properly at equilibrium or in an energy minimum, the first six modes will have frequencies of 0. These represent collective translations and rotations of the whole molecule along or about the three Cartesian axes and do not change the internal potential energy of the system (these can be eliminated beforehand by imposing the conservation of linear and angular momenta on the system). The vibrational modes are orthogonal or normal to one another, meaning that they can move independently. The excitation of one mode does not trigger the motion of a different mode. The general motion of the system can be described by a superposition of all modes. These normal modes yield analytical solutions to the equations of motion: for a given set of initial positions and velocities, NMA allows us to calculate where each atom of the system in question will be at any subsequent time subject to the small oscillation approximation.